SaaS M&A hit an all-time record in 2025 with nearly 2,700 transactions — up 28% year-over-year¹. Enterprise spending on generative AI grew more than 3x in 2025 alone, reaching $37 billion². And yet, the dominant narrative in our conversations this quarter is anxiety; a creeping fear that AI is about to detonate the entire enterprise software stack.
We understand the concern. It's not unfounded. AI agents are eliminating sticky learned interfaces. Business logic that once required top-tier engineering talent can now be approximated in a markdown file. Frontier models can query complex public datasets without a dedicated analytics product. Gartner³ estimates AI agents could replace 35% of point-product SaaS tools by 2030⁴ and a16z has noted that companies relying on vendor lock-in, or "hostages, not customers," will feel real pressure as switching friction erodes. These are significant disruptions, and some categories of software are genuinely at risk.
But panic is not a thesis. And notably, the loudest alarms are ringing at the top of the market, among the large incumbents whose moats were always more about inertia than innovation or irreplaceability. That's not where we invest.
At IGP, we've spent this quarter stress-testing our portfolio and deal pipeline through a simple but clarifying framework: real vs. perceived defensibility of moats against AI. What we're finding is that the AI wave is not a rising tide that capsizes all ships equally. It is separating the market into winners and losers with more precision than any prior technology cycle. In our opinion, the companies with risk share a common profile: their moat was never structural. Users stuck around because switching was inconvenient, not because the product was embedded in something irreplaceable. AI erodes process-based and habit-based friction. It does not erode embeddedness.
The companies we remain excited about are different. They sit on unique, proprietary data that frontier models can't access. They're deeply embedded in core, domain-specific workflows where AI adds efficiency without replacing the platform. They operate in compliant and regulatory-driven environments where "AI gained us 30% efficiency" is a feature, not a threat. They involve physical infrastructure or human-in-the-loop requirements that make substitution slow or legally prohibited.
In short: AI is an efficiency tailwind for the right businesses and an existential headwind for the wrong ones. The market is beginning to price this distinction. And beyond the benefits of scale efficiency, we also see innovation opportunity. When companies have to earn their customers' loyalty instead of relying on vendor lock-in, the result is faster innovation and a healthier competitive ecosystem for stars to rise.
While every company is unique and requires company-specific evaluation, we're currently focused on the following criteria in managing our portfolio and in sourcing. Not all companies will check every box, but many have multiple criteria:
- Proprietary Data. Companies that access unique, proprietary data that is not publicly accessible, which enables them to gain more value-creation leverage through AI — not despite it.
- AI-Forward or AI-Native. Built with and for AI advantages from the start.
- Compliant/Regulatory-Driven Requirements. Applications handling highly regulated and sensitive workflows, where trust, confidence, and legal responsibility are a must-have that cannot simply be replaced by an AI agent.
- Physical Infrastructure. Technology embedded within or connected to the physical world — extremely hard or impossible to disrupt with AI alone.
- Human-in-Loop Core Process. Critical business flows requiring human judgment — health services, legal determinations, physical tasks — where replacement is slow or prohibited.
- Domain Embeddedness. Software that is deeply embedded in core, domain-specific workflows, such as payments or cybersecurity, where the cost of extraction compounds over time.
Inside IGP, we also believe we have a strong advantage given our founding focus on tech-enabled operations. We built our CAPE platform — Comprehensive AI Process Enhancement. CAPE allows us to use first and third-party data across our organization to leverage AI, workflow automation, and insights. CAPE Sourcing uses engineering, data science, and LLM automation to find and engage deals, enabling what we believe is greater precision and speed than traditional outreach allows. It enables returned time for the senior relationships required to evaluate and close the best deals. We also recently launched CAPE Reporting to help automate portfolio company reporting and increasingly deploy it within deal and portfolio operations.
Our view heading into Q2: those best positioned for agility with defensible moats against AI can seize the opportunity. Criteria discipline will be critical, as investors continue to separate structural moats from legacy stickiness while evolving their own systems. However, the compression happening in undifferentiated SaaS is creating noise that masks genuinely durable assets trading at reasonable multiples. In our opinion, the opportunity isn't gone. It's just moved to higher ground.
Sources
- Gartner July 16, 2025 (ID G00834627)